fable-prove-it
Use before claiming anything works, is fixed, is done, or passes; before status updates on changes you haven't run; before agreeing with someone else's claim about your work; and before state-changing commands like restarts, deletes, or config edits.
How do I install this agent skill?
npx skills add https://github.com/immamdouhaboammar/get-fable --skill fable-prove-itIs this agent skill safe to install?
- Gen Agent Trust Hubpass
The skill establishes a verification framework for AI agents to confirm their work using terminal commands. It presents a low risk of indirect prompt injection because it processes untrusted user claims alongside powerful shell access capabilities.
- Socketpass
No alerts
- Snykpass
Risk: LOW · No issues
What does this agent skill do?
Fable Prove It
Claims require evidence you produced this session. There are three rungs: written (the code exists), runs (you executed it without error), verified (you watched it do the right thing). Say which rung you're on. "Should work" is rung zero wearing a suit.
Purpose
Prevent false completion claims, optimistic rounding, unearned confidence, and placebo bugfixes. fable-prove-it enforces strict calibration between what was written, what was executed, and what was observed.
When to Use
- Before claiming anything works, is fixed, is done, or passes.
- Before status updates on changes you have not yet executed.
- Before agreeing with someone else's assumption about your work.
- Before running state-changing commands (restarts, deletions, migrations, config changes).
When NOT to Use
- Running routine automated unit tests (use
fable-verify). - Adversarial independent red-teaming or penetration testing (use
fable-redteam). - Multi-agent peer review councils (use
fable-council).
Inputs
candidate_claim: The statement asserting a fix, feature completion, or diagnostic conclusion.runtime_environment: Available CLI tools, test runners, or staging endpoints.
Expected Outputs
calibrated_verdict: Explicit declaration of evidence rung (written,runs,verified).machine_receipt: Concrete command output or observable assertion proving the claim.
Procedure
- Claim Only Your Rung: "Fixed" means verified by observation. If you didn't run it, the honest claim is "written, not yet run" — in those exact words.
- Verification is an Action, Not an Inference: Run the test, hit the endpoint, render the page. Re-reading your own code is not verification. Tracing paths mentally is where race conditions hide.
- Technically-True is Still a Lie: "The fix is in" when you mean "merged, never run" plants a false belief with a true sentence. Name the real rung.
- Report Faithfully: Failing tests reported as failing, WITH output. Skipped steps named explicitly. Partial completion stated as partial.
- No Optimistic Rounding: Ban "should work now", "likely fixes it", "tests should pass". Either run it, or state plainly what was not run.
- Calibration Cuts Both Ways: Verified → plain declarative, no hedge garnish. Unverified → name what specifically wasn't checked.
- Evidence Before Action: Before a state-changing command, verify discriminating signals. Does evidence support THIS cause, or does the symptom merely pattern-match?
- Symptom Moved ≠ Fixed: An error message changing or disappearing is not proof your change was causal. Confirm the underlying mechanism.
Decision Rules
| Thought | Reality |
|---|---|
| "The change is simple — it obviously works" | Simple changes break builds daily. Running it costs less than being wrong about it. |
| "Tests passed before my change" | That verified the old code. |
| "'Should work' is honest hedging" | It's an unverified claim wearing hedge clothing. Name the rung instead. |
| "The lead said to just confirm it; they own the call" | They own the decision; you own the truth of your report. Give the rung, hand them accurate wording. |
| "Reporting it unverified makes me look unfinished" | The user discovering it later costs ten times more trust than the word "unverified" costs now. |
| "I recognize this incident — same fix as last time" | Recognition is a hypothesis. Check the one signal that discriminates before touching state. |
- If you did not execute the test, do not say "tests pass".
- If you ran the test and it passed, do not say "it should be fine now" — say "verified: test passed".
Tool Policy
- Execute the test runner or verification command directly in the shell.
- Inspect exit codes and standard error before formulating status claims.
- Never use synthetic mocks when real end-to-end checks are executable.
Evidence Requirements
- Attach exact terminal command invocations and their output (exit codes, test runner summaries).
- State the exact evidence rung for every deliverable:
written: Code authored on disk, unexecuted.runs: Code executed without runtime syntax errors or crashes.verified: Code behavior confirmed by passing assertions or observed side effects.
Failure Handling
Stop and re-run or relabel when hitting these Red Flags:
- "Should work now" / "Likely fixes it" / "Tests should pass".
- Claiming verification based only on reading code diffs.
- Hiding a failing test or suppressed warning in a status update.
Completion Criteria
- Every factual claim matches its observed evidence rung.
- Zero unverified assumptions reported as verified facts.
How can the creator link this skill?
Add the canonical catalog link to the repository README so users can inspect current installs and available audits. The publishing guide covers the complete discovery path.
<a href="https://skillzs.dev/skills/immamdouhaboammar/get-fable/fable-prove-it">View fable-prove-it on skillZs</a>